In [1]:
import os
import glob
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import plotly.express as px
import chart_studio.plotly as py
from IPython.display import IFrame
from datetime import datetime
from plotly.offline import init_notebook_mode, iplot
init_notebook_mode()
%matplotlib inline
In [2]:
confirmed_cases_path = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv"
deaths_path = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv"
cured_path = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv"
In [3]:
confirmed_cases = pd.read_csv(confirmed_cases_path)
confirmed_cases.head()
Out[3]:
In [4]:
deaths_data = pd.read_csv(deaths_path)
deaths_data.head()
Out[4]:
In [5]:
recovered_cases = pd.read_csv(cured_path)
recovered_cases.head()
Out[5]:
Let us see the cumulative growth of number of cases per each day over the entire world¶
In [6]:
days_columns = confirmed_cases.columns[4:]
In [7]:
world_cases_growth = confirmed_cases[days_columns].sum(axis = 0).reset_index()
world_cases_growth.columns = ['Date', 'Count']
In [8]:
def isweekend(date):
return datetime.isoweekday(pd.to_datetime(date)) > 5
world_cases_growth['isweekend'] = world_cases_growth['Date'].apply(isweekend).astype(int)
In [10]:
plt.rcParams['figure.figsize'] = [20, 10]
plt.stem(world_cases_growth['Date'], world_cases_growth['Count'], '--ro')
plt.plot(world_cases_growth['Date'], world_cases_growth['Count'], '--bo')
plt.title("Spread of virus per each day")
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.yticks(world_cases_growth['Count'])
plt.show();
Sudden spike can be seen from 12-Feb to 13-Feb almost 15000 cases were identified in just a single day . But most of them could be suspected cases.¶
Delta difference from day to day confirmed cases¶
In [11]:
world_cases_growth['delta_confirmed'] = world_cases_growth['Count'].sub(world_cases_growth['Count'].shift())
world_cases_growth.loc[0, 'delta_confirmed'] = world_cases_growth.loc[0, 'Count']
In [15]:
plt.rcParams['figure.figsize'] = [20,10]
plt.stem(world_cases_growth['Date'], world_cases_growth['delta_confirmed'], '--yo')
plt.plot(world_cases_growth['Date'], world_cases_growth['delta_confirmed'], '--ro')
for i in range(world_cases_growth.shape[0]):
plt.text(world_cases_growth.loc[i, 'Date'], world_cases_growth.loc[i, 'delta_confirmed']+2,
world_cases_growth.loc[i,'delta_confirmed'].astype(int))
plt.title("Delta new cases of to Covid-19 identified on each day")
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.show();
Parallelly let us compare the active cases to # of deaths on each day¶
In [16]:
world_cases_growth['deaths'] = deaths_data[days_columns].sum(axis = 0).reset_index(drop = True)
In [17]:
plt.rcParams['figure.figsize'] = [20,10]
plt.stem(world_cases_growth['Date'], world_cases_growth['deaths'], '--yo')
plt.plot(world_cases_growth['Date'], world_cases_growth['deaths'], '--ro')
for i in range(world_cases_growth.shape[0]):
plt.text(world_cases_growth.loc[i, 'Date'], world_cases_growth.loc[i, 'deaths']+50,
world_cases_growth.loc[i,'deaths'])
plt.title("Growth of death toll due to Covid-19 per each day")
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.show()
But the above representation is a cumulative sum of previous day count and newly added count, We can look at the delta change for every day.
In [18]:
world_cases_growth['delta_deaths'] = world_cases_growth['deaths'].sub(world_cases_growth['deaths'].shift())
world_cases_growth.loc[0, 'delta_deaths'] = world_cases_growth.loc[0, 'deaths']
In [19]:
plt.rcParams['figure.figsize'] = [20,10]
plt.stem(world_cases_growth['Date'], world_cases_growth['delta_deaths'], '--yo')
plt.plot(world_cases_growth['Date'], world_cases_growth['delta_deaths'], '--ro')
for i in range(world_cases_growth.shape[0]):
plt.text(world_cases_growth.loc[i, 'Date'], world_cases_growth.loc[i, 'delta_deaths']+2,
world_cases_growth.loc[i,'delta_deaths'].astype(int))
plt.title("# of deaths due to Covid-19 per each day")
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.show()
Let us also see how many cases were cured on daily basis¶
In [20]:
world_cases_growth['recovered'] = recovered_cases[days_columns].sum(axis = 0).reset_index(drop = True)
In [21]:
plt.rcParams['figure.figsize'] = [20,10]
plt.stem(world_cases_growth['Date'], world_cases_growth['recovered'], '--yo')
plt.plot(world_cases_growth['Date'], world_cases_growth['recovered'], '--go')
for i in range(world_cases_growth.shape[0]):
plt.text(world_cases_growth.loc[i, 'Date'], world_cases_growth.loc[i, 'recovered']+50,
world_cases_growth.loc[i,'recovered'])
plt.title("# of people that recovered from Covid-19 virus")
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.show()
In [25]:
plt.rcParams['figure.figsize'] = [20, 10]
plt.figure(1)
plt.subplot(311)
plt.plot(world_cases_growth['Date'], world_cases_growth['Count'], '--bo', label = 'ACTIVE CASES')
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.legend()
plt.subplot(312)
plt.plot(world_cases_growth['Date'], world_cases_growth['deaths'], '--ro', label = 'DEATHS')
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.legend()
plt.subplot(313)
plt.plot(world_cases_growth['Date'], world_cases_growth['recovered'], '--go', label = 'CURED')
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.legend()
plt.show()
Stacked Bar plot of confirmed, recovered and death counts on daily bases¶
In [27]:
p1 = plt.bar(world_cases_growth['Date'], world_cases_growth['Count'], color = 'yellow')
p2 = plt.bar(world_cases_growth['Date'], world_cases_growth['recovered'], color='g')
p3 = plt.bar(world_cases_growth['Date'], world_cases_growth['deaths'], color='r')
plt.xticks(world_cases_growth['Date'], rotation = 90)
plt.legend([p1[0], p2[0], p3[0]], ('Confirmed', 'Recovered', 'Deaths'))
plt.xlabel("Date")
plt.title("Stacked diagram of COVID-19 cases");
In [28]:
world_cases_growth['deaths'].values[-1]/world_cases_growth['Count'].values[-1]
Out[28]:
Increase in the trend of people recovering is a very good sign and considering the growth of confirmed cases and # of deaths, mortality rate is still as low as 4.6%¶
Let us now see the growth of virus in each country¶
In [29]:
con_cases_growth = confirmed_cases.groupby('Country/Region')[days_columns].agg(np.sum).reset_index()
con_cases_growth.head()
Out[29]:
As per the latest data lets see what is the severity of cases in each country¶
In [31]:
import requests
html_response = requests.get("https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_daily_reports?_pjax=%23js-repo-pjax-container")
In [39]:
from bs4 import BeautifulSoup
soup = BeautifulSoup(html_response.content, "html.parser")
In [43]:
a_tags = list(map(lambda x : x.get('href'), soup.findAll(name = 'a', attrs={'class' : 'js-navigation-open link-gray-dark'})))
In [45]:
repo_path = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_daily_reports/"+os.path.basename(a_tags[-2])
repo_path
Out[45]:
In [46]:
# daily_data = "csse_covid_19_data\\csse_covid_19_daily_reports"
# daily_data_dir = os.path.join(os.path.dirname(os.getcwd()), daily_data)
# daily_data_files = glob.glob(daily_data_dir+'/*.csv')[-1]
daily_data_files = repo_path
In [47]:
latest_data = pd.read_csv(daily_data_files)
# latest_data = latest_data.groupby(['Country_Region']).agg([])
latest_data.head()
Out[47]:
In [48]:
latest_data['Province_State'] = latest_data['Province_State'].fillna(latest_data['Country_Region'])
latest_data['Active'] = latest_data['Confirmed'] - (latest_data[['Deaths', 'Recovered']].sum(axis = 1))
In [49]:
agg_latest_data = latest_data.groupby('Country_Region')['Confirmed', 'Deaths', 'Recovered', 'Active'].agg([np.sum]).reset_index()
agg_latest_data.columns = agg_latest_data.columns.droplevel(1)
agg_latest_data.head()
Out[49]:
In [50]:
agg_latest_data = agg_latest_data.loc[(agg_latest_data[['Confirmed', 'Deaths', 'Recovered']] > 0).all(axis=1), :].reset_index(drop = True)
In [51]:
manual_dict = {'mainland china' : 'CHN', 'north macedonia' : 'MKD', 'palestine' : 'PSE',
'saint barthelemy' : 'FRA', 'south korea' : 'KOR', 'st. martin' : 'PYC',
'uk' : 'GBR', 'us' : 'USA', 'vatican city': 'ITA'}
def get_country_code(series):
df = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/2014_world_gdp_with_codes.csv')
df2 = pd.read_csv('countryCodes.csv', sep = '\t')
country_code_dict = {row[1][0].lower() : row[1][2] for row in df.iterrows()}
added_dict = {row[1][0].lower() : row[1][2] for row in df2.iterrows()}
# print(country_code_dict)
res = []
for key in series:
if key.lower() in country_code_dict:
res.append(country_code_dict.get(key.lower(), None))
elif key.lower() in added_dict:
res.append(added_dict.get(key.lower(), None))
else:
res.append(manual_dict.get(key.lower(), None))
return res
In [52]:
agg_latest_data['Code'] = get_country_code(agg_latest_data['Country_Region'])
In [53]:
from scipy.stats import rankdata
In [54]:
fig = px.scatter_geo(agg_latest_data, locations='Code', color=rankdata(agg_latest_data['Active'])/agg_latest_data.shape[0],
size = np.log(agg_latest_data['Active']+7),
hover_data = ['Country_Region', 'Active'],
projection="natural earth", color_continuous_scale = 'YlOrRd',
title="Count of Active COVID-19 cases in each country")
fig.update_geos(resolution=110,
showcountries=True, showcoastlines=False,)
# fig.show()
# fig.write_html("images/worldplot.html")
iplot(fig, filename='images/worldplot')
In [55]:
fig = px.choropleth(agg_latest_data, locations='Code', color=rankdata(agg_latest_data['Active'])/agg_latest_data.shape[0],
hover_data = ['Country_Region', 'Active'],
projection="natural earth", color_continuous_scale = 'YlOrRd',
title="Count of Active COVID-19 cases in each country")
fig.update_geos(resolution=110,
showcountries=True, showcoastlines=False,)
# fig.show()
# fig.write_html("images/worldplot.html")
iplot(fig, filename='images/worldplot')
On every day new people get affected, a part of affected people recover and some people die. Let us see what is the count of active cases in each day.
Province/State wise distribution of Active COVID-19 cases in Each country¶
In [56]:
latest_data.head()
Out[56]:
In [57]:
fig = px.treemap(latest_data, path = ['Country_Region', 'Province_State'], values = np.log(latest_data['Active']+7),
hover_data=['Active'], title = "Province/State wise distribution of Active COVID-19 cases in Each country")
fig.update_layout(width=2000,
height=1500)
iplot(fig, filename='images/treemap_active.html')
active = confirmed - (recovered + dead)¶
In [58]:
world_cases_growth['active_cases'] = world_cases_growth['Count'] - (world_cases_growth[['recovered', 'deaths']].sum(axis = 1))
In [59]:
fig = px.scatter(world_cases_growth, x = 'Date', y = 'active_cases', title='Daily count of active cases of COVID 19 through out the world')
# fig.show()
iplot(fig, filename='active_cases')
The number of Active cases have almost doubled with in the span of one week, 55K on 12-Mar and the number went past 120k on 18-Mar
Let us have a look at the list of countries in which Active cases grew. For now we will concentrate on a selected list of countries which have been appearing a lot in News.¶
In [81]:
confirmed_cases
Out[81]:
In [82]:
selected_countries = ['China', 'Italy', 'France', 'Spain','Germany', 'Iran',
'Korea, South', 'US', 'United Kingdom', 'Switzerland', 'India']
selected_dates = [1]+list(range(4, confirmed_cases.shape[1]))
# selected_dates = [1]+list(range(confirmed_cases.shape[1], confirmed_cases.shape[1]))
In [83]:
confirmed_cases_view = confirmed_cases.iloc[:, selected_dates]
confirmed_cases_view = confirmed_cases_view.loc[confirmed_cases_view['Country/Region'].isin(selected_countries)].groupby('Country/Region').agg([sum]).reset_index(drop=False)
confirmed_cases_view.columns = confirmed_cases_view.columns.droplevel(1)
confirmed_cases_view
Out[83]:
In [84]:
recovered_cases_view = recovered_cases.iloc[:, selected_dates]
recovered_cases_view = recovered_cases_view.loc[recovered_cases_view['Country/Region'].isin(selected_countries)].groupby('Country/Region').agg([sum]).reset_index(drop=False)
recovered_cases_view.columns = recovered_cases_view.columns.droplevel(1)
# recovered_cases_view
In [85]:
deaths_data_view = deaths_data.iloc[:, selected_dates]
deaths_data_view = deaths_data_view.loc[deaths_data_view['Country/Region'].isin(selected_countries)].groupby('Country/Region').agg([sum]).reset_index(drop=False)
deaths_data_view.columns = deaths_data_view.columns.droplevel(1)
# deaths_data_view
In [86]:
active_data_view = confirmed_cases_view.iloc[:, 1:] - (recovered_cases_view.iloc[:, 1:]+deaths_data_view.iloc[:, 1:])
# active_data_view
In [87]:
import plotly.graph_objects as go
In [89]:
fig = go.Figure()
for i in range(confirmed_cases_view.shape[0]):
fig.add_trace(go.Scatter(x = confirmed_cases_view.columns[1:], y = active_data_view.iloc[i, :],
name = confirmed_cases_view.iloc[i, 0]))
fig.update_layout(title = "Spread of COVID-19 in last 60 days in major Countries", height=800, width = 1000)
iplot(fig, filename='country_spread')
The graph shows how much a nation is prepared for an eidemic, clearly Italy was not expecting such massive outbreak. China on the otherhand showed the world that with proper care this Virus can be contained.
In [90]:
from ipywidgets import interact, widgets
from chart_studio.widgets import GraphWidget
In [91]:
@interact
def scatter_plot(country1 = countries, country2 = countries):
fig = go.Figure()
index1 = countries.index(country1)
index2 = countries.index(country2)
active_counts1 = active_data_view.iloc[index1, :]
active_counts2 = active_data_view.iloc[index2, :]
trace1 = fig.add_trace(go.Scatter(x = confirmed_cases_view.columns[1:], y = active_counts1, name=country1))
trace2 = fig.add_trace(go.Scatter(x = confirmed_cases_view.columns[1:], y = active_counts2, name=country2))
iplot(fig)